Papers by Marie-Francine Moens
Implicit Temporal Reasoning for Evidence-Based Fact-Checking (2023.findings-eacl)
Copied to clipboard
| Challenge: | Temporal reasoning is implicit since models learn from data how to leverage temporal information. |
| Approach: | They propose to ground claims and associated evidence on shared timelines using publication dates and time expressions extracted from their text. |
| Outcome: | The proposed model outperforms existing models that explicitly model temporal relations between evidence and the document by up to 9% Micro F1 and 15% Macro F1 on the MultiFC dataset. |
Mitigating Negative Interference in Multilingual Knowledge Editing through Null-Space Constraints (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing monolingual knowledge editing methods are expensive and require multiple models to maintain factual consistency. |
| Approach: | They propose a null-space constrained framework to precisely isolate language-specific knowledge updates that can be mapped onto other languages’ subspaces. |
| Outcome: | The proposed framework can project parameter updates for each language onto the orthogonal complement of other languages’ subspaces while preserving multilingual generalization capabilities. |
ClimateCause: Complex and Implicit Causal Structures in Climate Reports (2026.findings-acl)
Copied to clipboard
| Challenge: | Existing datasets for causal discovery from text lack granularity and abstraction for domains characterized by such complex causality. |
| Approach: | They propose a manually expert-annotated dataset of higher-order causal structures from science-for-policy climate reports. |
| Outcome: | The proposed dataset is highly readable and can be used to quantify readability. |
Fast-and-Frugal Text-Graph Transformers are Effective Link Predictors (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods that encode textual and structural information for inductive link prediction are frugal and fast at training and inference time. |
| Approach: | They propose a Transformer-based framework that unifies textual and structural information for inductive link prediction in text-attributed knowledge graphs by encoding ego-graphs (1-hop neighbourhoods). |
| Outcome: | The proposed framework can achieve superior performance on three popular datasets and reduce the reliance on resource-intensive encoders. |
A Flexible and Easy-to-use Semantic Role Labeling Framework for Different Languages (C18-2)
Copied to clipboard
| Challenge: | DAMESRL is an open source framework for deep semantic role labeling . language-specific characteristics and the available amount of training data influence the optimal model structure . |
| Approach: | They propose an open-source framework for deep semantic role labeling that is available under the Apache 2.0 license. |
| Outcome: | The proposed framework is available under the Apache 2.0 license. |
Generating Multiple-choice Questions for Medical Question Answering with Distractors and Cue-masking (2024.lrec-main)
Copied to clipboard
| Challenge: | Medical multiple-choice question answering (MCQA) requires high accuracy to be useful in practice. |
| Approach: | They propose to focus masked language modeling on disease name prediction when using medical encyclopedic paragraphs as input. |
| Outcome: | The proposed model outperforms the masked language model on disease name prediction and masks the cues to the answers. |
Language Fusion for Parameter-Efficient Cross-lingual Transfer (2025.acl-long)
Copied to clipboard
| Challenge: | Limited availability of multilingual text corpora for pretraining results in poor performance on downstream tasks due to undertrained representation spaces for languages other than English. |
| Approach: | They propose a method that integrates source and target language representations within low-rank (LoRA) adapters using lightweight linear transformations to enhance representation quality and transfer performance for languages other than English. |
| Outcome: | The proposed method improves representation quality and performance for languages other than English while maintaining parameter efficiency. |
CORE: A Few-Shot Company Relation Classification Dataset for Robust Domain Adaptation. (2023.emnlp-main)
Copied to clipboard
| Challenge: | Existing datasets that focus on company relations and business entities are lacking in relation classification. |
| Approach: | They introduce a few-shot relation classification dataset for company relations and business entities . they use a dataset that includes 4,708 instances of 12 relation types . |
| Outcome: | The proposed dataset includes 4,708 instances of 12 relation types with corresponding textual evidence extracted from company Wikipedia pages. |
Word-Level Loss Extensions for Neural Temporal Relation Classification (C18-1)
Copied to clipboard
| Challenge: | Unsupervised pre-trained word embeddings are used for many tasks in natural language processing to leverage unlabeled textual data. |
| Approach: | They extend the model's task loss with an unsupervised auxiliary loss on the word-embedding level of the model to ensure that the learned word representations contain both task-specific features and more general features. |
| Outcome: | The proposed model improves on the task of extracting narrative containment relations from clinical records using a general-domain part-of-speech tagger as linguistic resource. |
Augmenting BERT-style Models with Predictive Coding to Improve Discourse-level Representations (2021.emnlp-main)
Copied to clipboard
| Challenge: | Existing language models do not produce suitable representations at the discourse level. |
| Approach: | They propose to augment BERT-style language models with a mechanism that allows them to learn suitable discourse-level representations by incorporating top-down connections that operate at the intermediate layers of the network. |
| Outcome: | The proposed approach improves in 6 out of 11 tasks by detecting discourse relationship detection. |
Are Scene Graphs Good Enough to Improve Image Captioning? (2020.aacl-main)
Copied to clipboard
| Challenge: | Existing image captioning models rely on object detection features to generate image descriptions, but they are noisy. |
| Approach: | They propose to use scene graphs to introduce information about object relations into captioning to improve image descriptions. |
| Outcome: | The proposed model improves image caption quality by 3.3 CIDEr compared to a strong Bottom-Up Top-Down baseline. |
DMON: A Simple Yet Effective Approach for Argument Structure Learning (2024.lrec-main)
Copied to clipboard
| Challenge: | Argument structure learning (ASL) involves examining relationships between sentences in unstructured text. |
| Approach: | They propose a dual-tower multi-scale cOnvolution neural network to analyze relationships between arguments in a text. |
| Outcome: | The proposed approach outperforms state-of-the-art models on three domain argument mining datasets. |
Spatial and Temporal Language Understanding: Representation, Reasoning, and Grounding (2024.naacl-tutorials)
Copied to clipboard
| Challenge: | This tutorial provides an overview of cutting edge research on spatial and temporal language understanding. |
| Approach: | This tutorial provides an overview of cutting edge research on spatial and temporal language understanding. |
| Outcome: | This tutorial provides an overview of cutting edge research on spatial and temporal language understanding. |
Learning to Route for Dynamic Adapter Composition in Continual Learning with Language Models (2024.findings-emnlp)
Copied to clipboard
| Challenge: | Recent work shows that PEFT methods can be competitive with, or even superior to, full fine-tuning of PLMs. |
| Approach: | They propose a method that isolates the training of new PEFT modules to ensure their task specialization and learns to compose them by training a network of routers that leverages a small memory containing examples of previously seen tasks. |
| Outcome: | The proposed method improves generalization and performance in two CL setups. |
A Memory Model for Question Answering from Streaming Data Supported by Rehearsal and Anticipation of Coreference Information (2023.findings-acl)
Copied to clipboard
| Challenge: | Existing question answering methods assume that the input content can always be accessed while answering the question. |
| Approach: | They propose a model that performs rehearsal and anticipation while processing inputs to memorize important information for question answering tasks from streaming data. |
| Outcome: | The proposed model improves on short-sequence (bAbI) and large-squence textual (NarrativeQA) and video (ActivityNet-QA) question answering datasets. |
Representation, Learning and Reasoning on Spatial Language for Downstream NLP Tasks (2020.emnlp-tutorials)
Copied to clipboard
| Challenge: | In this tutorial, we discuss the cutting-edge research results and existing challenges related to spatial language understanding including semantic annotations, existing corpora, symbolic and sub-symbolic representations, qualitative spatial reasoning, spatial common sense, deep and structured learning models. |
| Approach: | This tutorial presents cutting-edge research results and current challenges related to spatial language understanding including semantic annotations, existing corpora, symbolic and sub-symbolic representations, qualitative spatial reasoning, spatial common sense, deep and structured learning models. |
| Outcome: | This paper reviews the cutting-edge research results and current challenges related to spatial language understanding including semantic annotations, existing corpora, symbolic and sub-symbolic representations, qualitative spatial reasoning, spatial common sense, deep and structured learning models. |
Talk2Car: Taking Control of Your Self-Driving Car (D19-1)
Copied to clipboard
| Challenge: | a long-term goal of artificial intelligence is to have an agent execute commands through natural language. |
| Approach: | They propose to use a dataset to compare commands written in natural language for self-driving cars with other datasets. |
| Outcome: | The proposed task is a challenging one and shows promising results, the authors argue . the talk2car dataset compares with similar datasets and shows that the proposed task requires additional research in natural language processing and computer vision. |
Visually-Aware Context Modeling for News Image Captioning (2024.naacl-long)
Copied to clipboard
| Challenge: | a new framework for News Image Captioning emphasizes the connection between textual context and visual elements. |
| Approach: | They propose a face-naming module for learning better name embeddings from news images . they use CLIP to retrieve sentences that are semantically close to the image . |
| Outcome: | The proposed framework outperforms the current state-of-the-art by 7.97/5.80 CIDEr scores on GoodNews/NYTimes800k. |
Finding Structural Knowledge in Multimodal-BERT (2022.acl-long)
Copied to clipboard
| Challenge: | Several multimodal-BERT models learn contextualized embeddings through training on linguistic data and visual data. |
| Approach: | They propose to make the structure of language and visuals explicit by a dependency parse . they also propose to encode the scene tree in the multimodal-BERT models . |
| Outcome: | The proposed models do not encode the scene trees in the language description. |
Evaluation Benchmarks for Spanish Sentence Representations (2022.lrec-1)
Copied to clipboard
Vladimir Araujo, Andrés Carvallo, Souvik Kundu, José Cañete, Marcelo Mendoza, Robert E. Mercer, Felipe Bravo-Marquez, Marie-Francine Moens, Alvaro Soto
| Challenge: | Existing and newly constructed datasets address different tasks from various domains. |
| Approach: | They propose to use Spanish SentEval and Spanish DiscoEval to evaluate stand-alone and discourse-aware sentence representations. |
| Outcome: | The proposed benchmarks evaluate the capabilities of stand-alone and discourse-aware sentence representations in Spanish and show that they are more robust and comparable than previous benchmarks. |
Efficient Information Extraction in Few-Shot Relation Classification through Contrastive Representation Learning (2024.naacl-short)
Copied to clipboard
| Challenge: | Relation classification (RC) models extract rich information from sentences with limited labeled instances. |
| Approach: | They propose to combine multiple sentence representations with contrastive learning to enhance information extraction by combining multiple sentence and entity tokens. |
| Outcome: | The proposed approach is able to extract discriminative information from multiple representations and contrastive learning. |
Temporal Information Extraction by Predicting Relative Time-lines (D18-1)
Copied to clipboard
| Challenge: | a new paradigm for temporal information extraction from text evades the relation extraction phase because there are n 2 possible entity pairs in a text with n temporal entities. |
| Approach: | They propose a method to construct a linear time-line from a set of temporal relations from text without the intermediate step of prediction of tempor relations. |
| Outcome: | The proposed method predicts start and end-points without intermediate step of prediction of temporal relations . it evades phase 2 because there are n 2 possible entity pairs in the extraction phase . |
Do Neural Network Cross-Modal Mappings Really Bridge Modalities? (P18-2)
Copied to clipboard
| Challenge: | Feed-forward networks are widely used in cross-modal applications to bridge modalities . success of such systems depends entirely on ability of mapping to make neighborhood structure akin to that of the target vectors. |
| Approach: | They propose to use a similarity measure to measure the neighborhood structure of neural network mappings. |
| Outcome: | The proposed model shows that the predicted neighborhood structure resembles more that of the input vectors than that of target vectors. |
Generating Continuous Representations of Medical Texts (N18-5)
Copied to clipboard
| Challenge: | Obtaining a good representation for medical texts may prove vital to building more sophisticated generative, discriminative or semantic models for the field. |
| Approach: | They propose an autoencoder that generates medical texts while learning an informative, continuous representation with discriminative features. |
| Outcome: | The proposed system generates medical texts while learning an informative, continuous representation with discriminative features. |
Computational Linguistics for Brain Encoding and Decoding: Principles, Practices and Beyond (2024.acl-tutorials)
Copied to clipboard
| Challenge: | This tutorial will explore the potential of computational linguistics to help understand brain language processing. |
| Approach: | This tutorial will explore the principles and practices of using computational linguistics methods for brain encoding and decoding. |
| Outcome: | This tutorial will explore the principles and practices of using computational linguistics methods for brain encoding and decoding. |
Modeling Coreference Relations in Visual Dialog (2021.eacl-main)
Copied to clipboard
| Challenge: | Visual dialog is a vision-language task where an agent needs to answer a series of questions grounded in an image based on the understanding of the dialog history and the image. |
| Approach: | They propose two soft constraints that can improve the model’s ability of resolving coreferences in dialog in an unsupervised way based on linguistic knowledge and discourse features of human dialog. |
| Outcome: | The proposed model achieves state-of-the-art performance on the VisDial v1.0 dataset without pretraining on other vision language datasets. |
Analysis and Prediction of NLP Models via Task Embeddings (2022.lrec-1)
Copied to clipboard
| Challenge: | Pretrained transformer-based encoders can be used to acquire rich text representations but need additional task supervision to be useful for downstream tasks. |
| Approach: | They propose a transformer to each MetaEval task and a neural network with a weighted encoder to perform the embeddings. |
| Outcome: | The proposed model outperforms baselines on GLUE tasks and can be used as a benchmark for future transfer learning research. |
Autoregressive Reasoning over Chains of Facts with Transformers (2020.coling-main)
Copied to clipboard
| Challenge: | Existing work on this task either evaluates facts in isolation or artificially limits the possible chains of facts, thus limiting multi-hop inference. |
| Approach: | They propose an iterative inference algorithm that decomposes the selection of facts from a corpus autoregressively and conditioning the next iteration on previously selected facts. |
| Outcome: | The proposed method outperforms the previous state-of-the-art in terms of precision, training time and inference efficiency on the WorldTree dataset. |
Learning Unsupervised Multilingual Word Embeddings with Incremental Multilingual Hubs (N19-1)
Copied to clipboard
| Challenge: | Recent research has found that a shared bilingual word embedding space can be induced by projecting monolingual word embeds from two languages without any bilingual supervision. |
| Approach: | They propose a framework for learning unsupervised multilingual word embeddings that mitigates instability issues for distant language pairs. |
| Outcome: | The proposed framework outperforms the state-of-the-art methods on two downstream tasks outperforming even supervised baselines. |
Decoding Language Spatial Relations to 2D Spatial Arrangements (2020.findings-emnlp)
Copied to clipboard
| Challenge: | Using a model architecture, we decode text to 2D spatial arrangements in a multi-object and multi-relationship setting. |
| Approach: | They propose a model architecture Spatial-Reasoning Bert that decodes language to 2D spatial arrangements in a multi-object and multi-relationship setting. |
| Outcome: | The proposed model can generate complete abstract scenes if paired with a clip-arts predictor and can generalize to out-of-sample data to a reasonable extent. |
Anti-Overestimation Dialogue Policy Learning for Task-Completion Dialogue System (2022.findings-naacl)
Copied to clipboard
| Challenge: | Recent research has focused on reinforcement learning (RL)-based dialogue policy. |
| Approach: | They propose a dynamic partial average estimator (DPAV) of the ground truth maximum action value to solve the overestimation problem. |
| Outcome: | The proposed method achieves better results on three dialogue datasets with a lower computational load compared to baselines on three different domains with lower bias. |
A Generic Method for Fine-grained Category Discovery in Natural Language Texts (2024.emnlp-main)
Copied to clipboard
| Challenge: | Existing methods for fine-grained category discovery neglect semantic similarities of fine-grain categories. |
| Approach: | They propose a method that detects fine-grained clusters of semantically similar texts guided by a novel objective function. |
| Outcome: | The proposed method surpasses state-of-the-art methods on three benchmark tasks. |
Sequence-to-Sequence Spanish Pre-trained Language Models (2024.lrec-main)
Copied to clipboard
| Challenge: | Spanish language models have demonstrated proficiency in natural language understanding and generation, but there is a scarcity of encoder-decoder models specifically designed for sequence-to-sequence tasks. |
| Approach: | They propose to implement encoder-decoder architectures pre-trained on Spanish corpora . they use them to assess sequence-to-sequence tasks including summarization, question answering . |
| Outcome: | The proposed models outperform models on sequence-to-sequence tasks in Spanish . the models show that they perform well across all tasks, the authors note . |
Protecting multimodal large language models against misleading visualizations (2026.acl-long)
Copied to clipboard
| Challenge: | MLLMs are robust to misleading visualizations, i.e., charts that distort the underlying data, leading readers to draw inaccurate conclusions. |
| Approach: | They propose to use table-based QA and redrawing the visualization to improve QA performance on misleading visualizations. |
| Outcome: | The proposed methods improve MLLM question-answering accuracy on misleading visualizations without compromising accuracy on non-misleading ones. |
“Image, Tell me your story!” Predicting the original meta-context of visual misinformation (2024.emnlp-main)
Copied to clipboard
| Challenge: | a surge of visual misinformation poses a growing threat to our society. |
| Approach: | They propose to use images to contextualize images to establish original meta-context . they use a framework that collects evidence and questions from the open web to ground the image . |
| Outcome: | The proposed approach helps human fact-checkers identify misinformation . it uses image contextualization to establish the original meta-context of the image . |
How Relevant is Selective Memory Population in Lifelong Language Learning? (2022.aacl-short)
Copied to clipboard
| Challenge: | Existing approaches to lifelong language learning rely on sparse experience replay to prevent catastrophic forgetting. |
| Approach: | They propose to use a selective memory population to store a uniform number of samples from the entire data stream to improve model performance. |
| Outcome: | The proposed methods show that they are relevant for lifelong language learning tasks, especially for low memory size, and consistent with computer vision studies. |